Scientific theories can't be causal arguments if they work just as well backwards in time. — Inis

Yes, causation can't be be backwards but a causal argument can refer to all phases of time, past, present, future.

Some empirical observations, such as the planet's in orbit round the sun were made before Newton formulated his theory of gravity. Some observations, like rockets following Newtonian mechanics, were made after. The point being a causal scientific theory can, or rather, should explain observed behavior, whether in the past, present or future.

Science is the Lying Game. A good lie must mix facts with the lie; and it must really hard to expose. In Sweden they give annually the prizes to the best lies of the year. Isaac Newton invented a lie that was only exposed four centuries later, that is why we consider him one of the greatest scientists in History. When you expose a great lie, you get to try to say another whopper; and Einstein took advantage of this rule to tell his own lies. They were so damn good they gave him the Swedish trophy as the best fabrication in Chemistry. Many physicists today dream of exposing Einstein´s relativistic lie; but it´s hard because Einstein was so good a concocting falsehood. Karl Popper helped to improve the Lying Game by introducing new rules. — DiegoT

Are you serious, or making a joke? Either way, it's funny.

The scientific method deals with universal statements — Inis

Does it always, though? Science does sometimes seem to concern itself with particular events, the conditions for which may not have been replicated at any other time or place (consider its studies of particular geological epochs, or particular events in geological history, such as the Permian-Triassic extinction).

Thanks for the replies. A lot to think over.

On the surface the Earth looks approximately flat. From a distance the Earth looks approximately spherical. That the Earth looks spherical is an observation. Yet we have examples of observations where things appear not quite as they really are, which we call optical illusions. There are optical illusions where straight lines appear curved (the straightness of the lines being judged in some other way).

So it could be that the Earth being spherical is an optical illusion. The statement that the Earth is spherical is based on the assumption that light travels in straight lines between the Earth and your eyes, or between the Earth and the camera taking pictures of the Earth from a distance. But the statement that light travels in straight lines is an untestable hypothesis, it is not falsifiable, because you never see light as it travels, you only see light when it reaches your eyes. It could be that light travels in such a way that Earth appears spherical while it is not. At first glance that seems to contradict other theories and observations, such as that gravity is spherically symmetric, but in fact you could formulate a theoretical framework in which Earth isn't spherical, in which gravity isn't spherically symmetric, in which light doesn't travel in straight lines, and which would fit observations just as well.

The equations in that theory would be probably more cumbersome, but then choosing between the two would just be a matter of convenience, fundamentally nothing makes the simpler theory more "true".

In fact you could formulate an arbitrarily large number of different theories that fit observations just as well. Most would be more complicated than what we have now. But potentially in the bunch lie some theories that are simpler than what we have now, and maybe we'll come to falsify theories which we treat as certainty now. But as Lakatos said, the act of falsification is nothing more than a social consensus, scientists decide that a theory is falsified when it encounters too many issues for their own taste and when other alternatives appear more fruitful.

A theory is basically a process that allows you to compute predictions from observations. But what lies beyond these observations is a matter of belief. We can tell how things appear to us, but we can't tell how they are independently of us, if that even has a meaning.

On the surface the Earth looks approximately flat. — leo

Except it doesn't when there is nothing in the way. Looking out to sea, or across a desert, the Earth looks curved. Then of course there is the phenomenon of the horizon etc.

But the statement that light travels in straight lines is an untestable hypothesis, it is not falsifiable, because you never see light as it travels, you only see light when it reaches your eyes. — leo

Except that we know that light doesn't travel in straight lines. We also know that in the time of Eratosthenes, it was an excellent approximation when he measured the circumference of the Earth 2.5k years ago.

It could be that light travels in such a way that Earth appears spherical while it is not. At first glance that seems to contradict other theories and observations, such as that gravity is spherically symmetric, but in fact you could formulate a theoretical framework in which Earth isn't spherical, in which gravity isn't spherically symmetric, in which light doesn't travel in straight lines, and which would fit observations just as well. — leo

All observations are theory-laden. There is no escaping that. However, claiming that there could be some theoretical framework which is in better accordance with reality, while simultaneously rendering a sequence of optical illusions, simply won't do. You have to come up with the actual theory, so that we can come up with some tests to decide which theory wins.

You don't understand that the trajectory of light is pure assumption, it cannot be observed directly. There are theories where light travels in straight lines, there are theories where it doesn't, both can be made to fit with observations. As Popper and Lakatos explained you can always save whatever non-trivial theory by adding auxiliary assumptions to it. You can always assume there is some additional effect not taken into account acting on what is measured or on the instruments of measurement themselves. At any time there are an arbitrarily large number of theories consistent with a finite number of observations. Besides, by some mathematical transformation you can reformulate a theory in a different geometrical space such that things have a different shape while the theory remains empirically equivalent. I wouldn't know how to make you see that in a few lines if you don't see it, but start by trying to see why you can't observe the trajectory that light follows, you only see light reaching your eyes. Then you can assume different trajectories and reformulate the 'laws' of physics such that they fit these trajectories, and then things such as the Earth would have a different shape than how they appear.